Effectively inferring overall spatial distribution pattern of species in a map when exact coordinate information is missing

Abstract

Abstract Mapping and comparing biodiversity in space is a routine task in ecological research. However, accurate biodiversity mapping requires exact distributional information as inputs. When exact geographical coordinate information is missing or when only coarse distribution of species is available at the quadrat or plot level, how can we effectively infer and compare spatial distributional patterns of species with reasonable accuracy while controlling the confounding impacts of scale dependency? A crossing‐scale coefficient of variation (CCV) is calculated for three simple diversity indices, the Shannon, Gini–Simpson and reciprocal‐Simpson evenness indices, from the quadrat‐based biodiversity data over varying sampling grain sizes for subtly measuring overall spatial distributional pattern of species in a survey map. Extensive numerical simulations showed that, when the overall distribution of organisms was regular, CCV tended to be very small, for all the three evenness indices. By contrast, when the distribution was highly aggregate, CCV tended to be large. Finally, an intermediate CCV was observed when organism distribution was random. To this end, all the three evenness indices can detect the subtle change on aggregate or regular distribution. Among the indices, the Gini–Simpson evenness index is the most insensitive to both the change of population size of species and the number of sampling grain sizes used. Compared to available spatial statistic metrics that requires exact geographical coordinates as inputs, CCV is a useful metric for assessing the overall spatial distribution pattern of a species across sampling plots in which exact spatial coordinates of each individual are unavailable. Empirical applications of the CCV metric showed that plants in habitat patches extensively managed by humans can present very aggregate distributional pattern, being counterintuitive to the usual expectation that plants in cultivated or planted land should present regular distributional pattern. The present study recommends one effective tool, the combination of CCV and evenness indices, for exploring the overall spatial distributional pattern of a single species under the quadrat‐sampling scheme. The proposed method is particularly suitable to handle ecological data for which the finest‐scale distributional coordinate information of each individual of species is missing.